Oncogene www.nature.com/onc ARTICLE OPEN Intratumor genetic heterogeneity and clonal evolution to decode endometrial cancer progression Alba Mota1,2,14, Sara S. Oltra1,2,3,14, Pier Selenica4, Cristian P. Moiola3,5, Carlos Casas-Arozamena 6, Carlos López-Gil5, Eva Diaz1, Sonia Gatius7, María Ruiz-Miro8, Ana Calvo9, Alejandro Rojo-Sebastián3,5,10, Pablo Hurtado11, Roberto Piñeiro 11, Eva Colas3,5, Antonio Gil-Moreno 3,5,12, Jorge S. Reis-Filho 4, Laura Muinelo-Romay3,6, Miguel Abal 3,6, Xavier Matias-Guiu3,7,13, ✉ Britta Weigelt4 and Gema Moreno-Bueno 1,2,3 © The Author(s) 2022 Analyzing different tumor regions by next generation sequencing allows the assessment of intratumor genetic heterogeneity (ITGH), a phenomenon that has been studied widely in some tumor types but has been less well explored in endometrial carcinoma (EC). In this study, we sought to characterize the spatial and temporal heterogeneity of 9 different ECs using whole-exome sequencing, and by performing targeted sequencing validation of the 42 primary tumor regions and 30 metastatic samples analyzed. In addition, copy number alterations of serous carcinomas were assessed by comparative genomic hybridization arrays. From the somatic mutations, identified by whole-exome sequencing, 532 were validated by targeted sequencing. Based on these 1234567890();,: data, the phylogenetic tree reconstructed for each case allowed us to establish the tumors’ evolution and correlate this to tumor progression, prognosis, and the presence of recurrent disease. Moreover, we studied the genetic landscape of an ambiguous EC and the molecular profile obtained was used to guide the selection of a potential personalized therapy for this patient, which was subsequently validated by preclinical testing in patient-derived xenograft models. Overall, our study reveals the impact of analyzing different tumor regions to decipher the ITGH in ECs, which could help make the best treatment decision. Oncogene (2022) 41:1835–1850; https://doi.org/10.1038/s41388-022-02221-0 INTRODUCTION ambiguous EC (AEC) has mixed, overlapping features or lack of Endometrial cancer (EC) is the most common gynecological cancer any evidence of differentiation at both the microscopic and in developed countries, and it accounts for nearly 5% of cancer molecular levels [5–7], and it may not always fit into the molecular cases and more than 2% of cancer-associated deaths in women TCGA subtypes or the traditional dualistic histologic classification worldwide [1]. EC has traditionally been classified into two main [5, 6]. These tumors have been previously described by other groups, type I and type II carcinomas, with different endocrine, authors and account for less than 1% of all ECs [5–7]. They are clinical, pathological, and molecular features [2, 3]. More recently, often associated with aggressive behavior [7] and with the lack of The Cancer Genome Atlas (TCGA) defined a molecular classifica- response to conventional therapies [8]. The diagnosis of AEC is tion of EC [4] based on somatic mutations, copy number difficult to make in the absence of an intense molecular analysis, alterations and microsatellite instability status. Four distinct EC as performed in this study. AEC cases are occasionally placed in molecular subtypes were established, associated with distinct the spectrum of high-grade endometrioid carcinomas, although outcomes, namely the POLE/ultramutated, microsatellite instable they are very aggressive, and do not fit perfectly in any TCGA (MSI)/hypermutated, copy-number low (CN-low)/endometrioid subgroup. For these reasons, it is important to be aware of the and copy-number high (CN-high)/serous-like subtypes. Although existence of this unusual type of tumor. this TCGA study improved the existing classification of ECs, there Intratumor genetic heterogeneity (ITGH) has emerged based on are still rare EC histologies that are challenging to classify. There is the analyses of different areas of the primary tumor and metastatic a ‘grey zone’ between grade 3 endometrioid and serous lesions from the same patient using whole-exome sequencing carcinomas, with some cases exhibiting overlapping morphologi- (WES) and targeted massively parallel sequencing [9–12]. In this cal and molecular features [5, 6]. For example, high-grade regard, we detected strong ITGH in primary ECs [13, 14], as 1 MD Anderson International Foundation, 28033 Madrid, Spain. 2Biochemistry Department, Universidad Autónoma de Madrid (UAM), Instituto de Investigaciones Biomédicas ‘Alberto Sols’ (CSIC-UAM), IdiPaz, 28029 Madrid, Spain. 3Centro de Investigación Biomédica en Red de Cáncer (CIBERONC), 28029 Madrid, Spain. 4Department of Pathology, Memorial Sloan Kettering Cancer Center, New York, NY 10065, USA. 5Biomedical Research Group in Gynecology, Vall Hebron Institute of Research, Universitat Autònoma de Barcelona, 08035 Barcelona, Spain. 6Translational Medical Oncology Group (Oncomet), Health Research Institute of Santiago de Compostela (IDIS), University Hospital of Santiago de Compostela (SERGAS), Trav. Choupana s/n, 15706 Santiago de Compostela, Spain. 7Department of Pathology, Hospital U Arnau de Vilanova, University of Lleida, IRBLLEIDA, Lleida, Spain. 8Biobank, IRBLLEIDA, Lleida, Spain. 9Department of Gynecology, Hospital U Arnau de Vilanova, IRBLLEIDA, Lleida, Spain. 10MD Anderson Cancer Center, Madrid, Spain. 11Roche-Chus Joint Unit, Translational Medical Oncology Group (Oncomet), Health Research Institute of Santiago de Compostela (IDIS), Travesía da Choupana s/n, 15706 Santiago de Compostela, Spain. 12Gynaecological Department, Vall Hebron University Hospital, 08035 Barcelona, Spain. 13Departments of Pathology, Hospital U. de Bellvitge, Universities of Lleida and Barcelona, IDIBELL Lleida and Barcelona, Spain. 14These authors contributed equally: Alba Mota, Sara S Oltra. ✉email: gmoreno@iib.uam.es Received: 27 May 2021 Revised: 12 January 2022 Accepted: 27 January 2022 Published online: 10 February 2022 A. Mota et al. 1836 OS (years) previously described for other tumor types [15]. However, these studies did not address the spatial and temporal genetic 10.18 10.37 6.73 12.73 1.57 1.51 2.32 2.25 heterogeneity among ECs, nor that between ECs and their distant 10.9 metastases. Notably, ITGH analysis allows for the identification of sub-clonal variants present at low frequencies and non-uniform distribution across tumor regions. These variants could be (years) important elements in tumor evolution and subsequently, in the 3.12 0.26 1.21 7.81 0.21 0.49 0.97 RFS 3.1 1.4 patient’s response to treatment and for their prognosis [11, 16–18]. To get deeper insights into the behavior of ECs and to be able Statusb to adapt the clinical management of EC patients accordingly, it is AWD AWD DOD DOD DOD DOD DOC imperative to decode the genomic evolution of these tumors. AW DF Thus, we sought here to define the spatial and, in some cases, temporal heterogeneity of ECs, their clonal composition, and the Peritoneum Peritoneum Diaphragm Metastasis clonal shifts between primary and metastatic disease. Spatially Iliac node distinct samples of primary tumors and metastatic lesions from Ovary Ovary Ovary Ovary Node five patients with metastatic endometrioid EC (EEC; three of which were MSI), three patients with metastatic serous EC (SECs), and one patient with high-grade AEC, were subjected to WES, targeted massively parallel sequencing, and a subset to array comparative classification genomic hybridization (aCGH) to decipher their genomic land- TP53 IHC: 2+ = p53 overexpression, indication of missense mutation; 1+ = wild-type pattern; 0 = absence of protein, indication of null mutation. Serous-like Serous-like Serous-like Serous-like scape. These data were employed to reconstruct the spatial and CN-low CN-low temporal, when was possible, evolution of the tumors within each TCGA patient. Several phylogenetic patterns were identified in these MSI MSI MSI tumors, independent of their classic histological or molecular classification, including similar patterns in cases with ovarian metastasis or recurrent disease. Despite the abundant spatial ITGH TP53 IHCa within the primary ECs, the majority of anatomically distinct EC 1+ 1+ 2+ 1+ 1+ 2+ 2+ 1+ PDX patient-derived xenograft, WES whole-exome sequencing, CNA copy number aberration, P primary tumor, M metastasis. 0 metastatic lesions from a given patient displayed genomic homogeneity. In addition, we defined the molecular evolution in Positive-high Positive-high Positive-high Positive-low an AEC tumor. Interestingly, our analyses on AEC identified new MSI status Negative Negative Negative Negative Negative potential targetable biomarkers that were validated by treating a patient-derived xenograft (PDX) with specific therapies. Overall, our results could help to prompt new paths for the personalized Clinicopathological characteristics of the endometrial cancer patients included in the study. AW alive and well, AWD alive with disease, DF disease free, DOC died of other cause, DOD died of disease. treatment of EC patients and potentially for a rare subtype of these cancers like AEC. POLE status RESULTS WT WT WT WT WT WT WT WT WT Clinicopathological characteristics of ECs included in this study Lymph node Nine metastatic EC patients with complete clinical follow-up information were included in this study: 5 endometrioid ECs Yes (11/36) invasion (EEC3-7), 3 serous ECs (SEC1-3) and 1 AEC. Immunohistochemistry (IHC) analysis of the DNA mismatch repair (MMR) proteins and p53 Yes No No No No No – – (Supplementary Fig. 1A), combined with the POLE sequencing results (Table 1), were employed to define the ProMisE molecular All patients received the conventional treatment caboplatin-Taxol. subtypes [19, 20]. This analysis revealed that none of the nine ECs included in this study harbored a POLE hotspot mutation (data not Myometrial infiltration shown), that 3/5 EECs lacked MLH1 and PMS2 expression and were classified as MSI molecular subtype, that 2/5 EECs were classified <50% <50% 100% 80% 40% 75% 95% 80% 95% as CN-low, and that all 3 SECs showed aberrant p53 expression and were classified as CN-high (serous-like) (Table 1). The AEC tumor displayed ambiguous histologic characteristics FIGO and when the ProMisE molecular surrogate was applied, the IIIA IIIA IIIC IIIB IA IV IB IB IB tumor was classified as CN-low or as a tumor with a non-specific molecular profile [20, 21]. Moreover, conventional histopathologi- Grade cal and IHC assessment defined a high-grade carcinoma that did not fulfill the typical features of an endometrioid, serous, clear cell, 2 2 3 3 3 3 3 3 3 or undifferentiated carcinoma (Supplementary Fig. 1A). After Endometrioid Endometrioid Endometrioid Endometrioid Endometrioid having been assessed by two blinded expert gynecological Ambiguous pathologists, the high confidence score (0.548) obtained in the Histology TumorTracer study [22] confirmed the endometrial origin of this Serous Serous Serous tumor by comparing the mutational profile of our AEC with that of more than 7600 tumors (Supplementary Fig. 1B). This tumor exhibited a solid pattern of growth with an absolute lack of any Patient Table 1. type of differentiation (endometrioid, serous, clear cell). In SEC-1 SEC-2 SEC-3 EEC-3 EEC-4 EEC-5 EEC-6 EEC-7 AEC addition, although AEC could misclassify with undifferentiated b a carcinoma, did not show any of the molecular and Oncogene (2022) 41:1835 – 1850 A. Mota et al. 1837 immunohistochemical features of these tumors [23]. These data serous-like samples had an aging-related signature 1, whereas all provide further evidence that the tumor has an endometrial origin MSI tumors displayed signatures 6 or 20, linked to deficient DNA as opposed to metastasis of an extrauterine malignancy. MMR. Additionally, HPV testing was also negative, ruling out an origin in the uterine cervix (Supplementary Fig. 1C). In light of the complexity of the AEC tumor and to corroborate its phenotype, an CLONALITY ANALYSIS REVEALS A MAINLY MONOPHYLETIC additional IHC study was performed with a panel of biomarkers EVOLUTION OF METASTATIC ECS previously used to differentiate between high-grade endometrioid A clonal study was carried out to further assess the evolution for and serous carcinoma (see Fig. 6A, for more details) [24]. This each tumor, which revealed important differences in the analysis was inconclusive for all the tumor areas tested and thus, proportion of sub-clonal mutations between EECs and SECs the tumor was classified as a non-specific AEC tumor. Indeed, (unpaired t-test, P = 0.02: Supplementary Fig. 2C). All except two based on the gross, microscopic and IHC data, this tumor was EECs (5/7, 71%) showed less than 10% of sub-clonal variants confirmed as a high-grade EC with ambiguous features, FIGO (except for EEC3 and EEC7 with 16% and 20%, respectively), whilst stage IIIC2. SECs presented a mean of 27% sub-clonal variants. Furthermore, Although small sample size, comparison based on TCGA to gain further insight into tumor evolution, phylogenetic trees classification (including CN-low n = 2, MSI n = 3, serous n = 3 were generated based on the clusters identified by PyClone [30]. and AEC n = 1) and histological type (Endometrioid n = 5, Serous Interestingly, and as described previously [27], hierarchical n = 3 and AEC n = 1) revealed significantly different overall clustering showed a main pattern of monophyletic evolution, survival rates (Supplementary Fig. 1D, p = 0.025 and p = 0.01, with primary tumor regions more closely related among respectively). Interestingly, the case of AEC and serous tumors themselves than with the metastatic lesions. Metastatic lesions were associated with the poorest survival. Even though, these from several patients seem to arise from an ancestor clone (EEC3, results should be confirmed in a larger cohort, our analysis EEC4, EEC6, EEC7, SEC1 and SEC3), with both primary and supported previous studies that showed an association of metastatic tumors also acquiring subsequent mutations (Fig. 2 endometrioid cases with low-intermediate-risk and serous-like and Supplementary Fig. 2D–F). In the CN-low subgroup (EEC3 and groups with the worst outcomes [25]. EEC4: Fig. 2A, B), a main ancestor clone represented 61.2% and 40.1% of the common variants between primary tumor (T1 and T2) and metastatic (M) regions, respectively. Of note, different THE MUTATIONAL REPERTOIRE IDENTIFIED BY WES REVEALS regions of CN-low tumors were very similar among themselves. A PROFILE CONSISTENT WITH EECS AND SECS By contrast, EEC5 (Fig. 2C) followed a polyphyly evolution, being A total of 40 frozen samples from these 9 patients (5 EEC, 3 SEC, the metastatic sample arising from a specific tumor region. In this and 1 AEC) were analyzed by WES to a 52× median depth (range case, just 9.78% of the variants were common to the T1, T2 and M = 30–70×), recording the somatic variants identified in each regions, characterized by signatures 1 and 6. In the course of sample (Supplementary Table S1). A median of 335.4 (range = tumor evolution, the metastatic region was phylogenetically closer 24–1537) somatic variants and 44.3 (range = 1–213) insertions/ to T2, sharing a subset of 334 variants (16.4%), in contrast to T1 deletions (indels) per sample were identified, with a median of and M, in which none of the variants were exclusively shared by 179.6 (range = 12–814) non-synonymous somatic variants (Fig. 1A them. Indeed, the primary T1 tumor region acquired different and Supplementary Table S2). These results were similar to those exclusive mutations (33.4%), with an increase in the representa- reported in the TCGA [26]. The WES data were in accordance with tion of signature 26 in the final mutations gained (also related to the IHC studies and confirmed the molecular classification of EEC3 defective DNA MMR pathway). and EEC4 as CN-low; EEC5, EEC6 and EEC7 as MSI; and SEC1, SEC2 Despite the polyphyly evident in EEC5, the EEC6 MSI patient and SEC3 being CN-high. The AEC classification was controversial (Fig. 2D) exhibited a monophyletic evolution in which the main and is discussed in more detail below. As expected, samples from ancestral clone contributed 18.8% of the common variants ECs of MSI subtype harbored significantly more (One-way ANOVA between the T1 or T2 and the M regions. Regarding the mutations test P < 0.001) genetic variants (median = 835.4, range = identified, 32.9% of them were exclusively acquired in the 283–1537) than samples from ECs of CN-low (median = 226.16, metastatic region, while 11.6 % and 22.1% were acquired range = 60–404), and CN-high subtypes (median = 63.3, range = separately by the primary T1 and T2 regions, respectively. A 24–112) and the AEC case (median = 233.3, range = 196–253: Fig. distinct progression was observed in EEC7 (Fig. 2E), in which the 1B). No significant differences in the number of mutations ancestor clone only shared 1.46% of the variants in the three (synonymous and non-synonymous) were detected between tumor regions. In this sub-clone, the primary tumor regions, and paired primary tumor and metastases samples (Supplementary the metastatic lesion (a recurrent disease diagnosed 7 years after Fig. 2A). Furthermore, all samples analyzed harbored mutations in the primary tumor) followed a differential progression, with 45.8% more than one of the most significantly mutated genes in ECs of the mutations identified coinciding in the T1 and T2 regions. reported by the TCGA [26, 27] (Fig. 1C), with the exception of the Mutational signatures suggest high diversity in their tumor AEC case, which only harbored a mutation in the ESR1 gene. Thus, evolution, being signature 20 the main one in primary tumor the most frequently mutated genes in this series were PIK3CA (5/9 samples and signature 6 that of the metastasis (both related to patients, 4 EECs, 1 SEC), PTEN (5/9 patients, all EECs), and as DNA MMR deficiency) [28]. These results might reflect the early expected, TP53 and PPP2R1A were mutated in the SECs (Fig. 1C). progression of an initial tumor sub-clone with stem properties that In accordance with the WES results, we observed significantly would have given rise to the recurrent disease. However, more common SNVs shared by all the tumor regions in the CN-low additional regions from this patient should be analyzed to confirm subgroup (EEC3 and EEC4), which shared more than 40% of the the distinct progression observed in this case. Alternatively, this somatic variants detected (Supplementary Fig. 2B). By contrast, recurrent disease could reflect clonal reduction caused by the the MSI subgroup shared fewer SNVs (19%), suggesting an received treatment. Nevertheless, no results relative to therapy increased ITGH in this subgroup. An intermediate ITGH was could be extracted due to all patients received the conventional observed for the CN-high SECs, whereby the number of common Carboplatin-Taxol therapy, therefore further studies should be somatic variants in all the samples analyzed was above 26% (SNV addressed to explore this issue. range = 26.8–41.2%; Supplementary Fig. 2B). Regarding the serous tumors, phylogenetic analysis also We also interrogated the mutational signatures in our series revealed a monophyly evolution and in all cases, the number of [28, 29] (Fig. 2 and Supplementary Fig. 2). All CN-low and CN-high/ variants shared by all the regions analyzed was above 20%, Oncogene (2022) 41:1835 – 1850 A. Mota et al. 1838 Fig. 1 Immunohistochemical and genomic representations of endometrial carcinomas. A Graphical representation of the somatic variants (SNVs) detected by samples analysis and according to the endometrial molecular subgroup (B). C Representation of a selection of the most common pathogenic somatic mutations described previously in endometrial carcinomas [26, 27]. Mutation types are colored according to the legend and the endometrial subgroups are indicated: CN copy number, MSI microsatellite instability, AEC ambiguous endometrial carcinoma. The presence of different mutation types for each EC patient were grouped for primary tumors (P) and metastatic regions (M). indicative of lower levels of ITGH. In SEC1 (Supplementary Fig. 2D), amplicon sequencing to overcome these problems and quantify 26.8% of the variants identified in the ancestor clone were shared the true ITGH. by the metastatic and primary tumor regions. During tumor evolution, the metastatic region retained 7.04% of the genetic variants found and primary tumor regions were grouped in a main DECODING THE ITGH IN EC BY TARGETED MASSIVELY branch. SEC2 (Supplementary Fig. 2E) was composed of a major PARALLEL SEQUENCING clone in which the 30.8% of the variants were common to the AmpliSeq analysis was performed to validate the mutations three primary tumor regions and the metastatic region. From this identified by WES and to perform a more detailed characterization clone, each tumor lesion evolved and acquired a specific of the heterogeneity among the samples from the same patients. mutational pattern, all of which were enriched for C > T transitions Targeted massively parallel sequencing was performed on all in the NpCpG context, consistent with signature 1 and associated samples analyzed by WES and additional 41 formalin fixed paraffin with aging [28]. Similar results were obtained for the third SEC embedded (FFPE) samples (20 from metastasis and 21 from (SEC3, Supplementary Fig. 2F), which presented the strongest primary tumor) (Supplementary Table S3). A total of 532 variants similarity in the four tumor regions analyzed (41.2% of common were reanalyzed in the validation process (Supplementary Table variants between T1, T2, M1 and M2). S4). The selection criteria included all the pathogenic variants Notably, we obtained a landscape of the EC mutation profile along with the additional non-pathogenic variants, reaching at from the WES analysis and the evolution of the regions analyzed. least 40% of those detected by WES in the CN-low, CN-high and Moreover, WES allowed us to identify the molecular subtypes in AEC case, and 10% in the MSI ECs. each sample, their associated mutational signatures, and their Among the variants included in the validation study, we temporal evolution. However, the low sequencing depth in this compared the proportions of somatic mutations shared by all study and the small proportion of tumor samples impaired the the samples in the WES study relative to the validation targets detection of low-frequency mutations that often account for the (Supplementary Fig. 3A). For the CN-low EC, the CN-high SECs and majority of the ITGH in a cancer. Moreover, WES studies may the AEC, the number of shared variants detected in the targeted produce confounding results by overestimating the ITGH [18]. sequencing was similar or fewer than the common variants in the Thus, the mutational landscape defined in the WES study could WES study (Supplementary Fig. 3A). For the MSI ECs, an increase in help to select the most interesting variations in a tumor, although the shared variants in the targeted sequencing was found relative it will be necessary to further validate these through high-depth to the WES analyses (Supplementary Fig. 3A). Indeed, the ITGH Oncogene (2022) 41:1835 – 1850 A. Mota et al. 1839 Fig. 2 Tumor evolution of endometrial carcinomas. Venn diagrams representing the somatic mutations shared by the primary tumor sections and metastases analyzed by WES. The numbers of the genetic variants are shown, and the percentages are in brackets. Phylogenetic trees based on somatic mutations depicting the evolution of two primary tumor areas and a metastasis. The length of the branches is proportional to the number of shared or private somatic mutations. The pie charts represent the mutational signatures previously described [29] for the shared or private somatic mutations: A EEC3, B EEC4, C EEC5, D EEC6, E EEC7. WES whole-exome sequencing, EEC endometrioid endometrial carcinoma. Oncogene (2022) 41:1835 – 1850 A. Mota et al. 1840 rate was less pronounced in the targeted sequencing analysis for an early spread of the metastatic clone to the ovary and the the MSI tumors. Probably, this is due to the fact that the subsequent development of the primary tumor in the uterus sequencing depth was significantly higher in targeted sequencing (Fig. 4A, B and Supplementary Fig. 3B). Thus, in SEC1, a group of than in the WES analysis (more than 20-fold) even allowing to mutations were mostly identified in the primary tumor regions better capture the low-frequency variants detected in the (Fig. 4A). A private mutation in the DCP1B gene of the WES study. metastasis M1 region and a metastasis exclusive mutation in The MSI ECs exhibited higher levels of ITGH among the regions PXDNL were identified in SEC2 (Fig. 4B, Supplementary Table analyzed and the targeted sequencing provided more information S4). In SEC3, a mutation in TRIM27 was restricted to M2 due to the improved depth of coverage. In case EEC5, only 17,93% (Supplementary Fig. 3B, Supplementary Table S4). Together, (33/184) of the genetic alterations selected for validation were these results suggest a clonal independent synchronous found in all the samples analyzed by WES (Supplementary Fig. 3A). evolution for the primary and metastatic regions of SECs. However, this percentage increased in the validation study, with However, the time of sample collection was the same for 35,8% of the variants found in all the samples analyzed by primary tumor and metastasis regions and this represents a targeted sequencing. This increase highlights the potential of the limitation, being required the analysis of additional samples to targeted sequencing to detect low-frequency variants that were confirm the trend. Regarding CN-low cases, more than 70% of missed by WES. Indeed, 13.6% of the variants included in the the variants were identified in the primary tumor and validation study were restricted to the primary tumor samples by metastatic samples, (Supplementary Fig. 3C, D, Supplementary WES (identified in EEC5 T1 and/or T2), yet they were subsequently Table S4). In both cases, the metastatic regions analyzed found in the metastatic regions by targeted sequencing. Despite belonged to a recurrent disease, affecting lymph nodes and the an increase in the variants detected, the targeted validation study peritoneum in EEC3 and EEC4 patients, respectively. This is revealed ITGH among the different regions analyzed, as was reflected in the phylogenetic trees with a long branch of evident through the irregular distribution of the variants ubiquitous shared variants and a final expansion of the identified. ITGH was even found in hotspot mutations, affecting metastatic regions due to the acquisition of additional driver genes or genes that could potentially be implicated in mutations (Supplementary Fig. 3C, D bottom). treatment selection, including TP53 p.G245S and p.G244C muta- tions (labeled in red: Fig. 3A, Supplementary Table S4). The phylogenetic tree of this tumor reflected the ITGH, with multiple HIGH SOMATIC COPY NUMBER ALTERATIONS ITGH IN SECS branches and sub-clones, suggesting an evolution of the In contrast to the somatic mutations, widespread heterogeneity of metastatic sample from the primary T3 tumor region (Fig. 3A somatic copy number alterations (SCNAs), defined by aCGH bottom), and an indication of polyphyly. analysis, was found in the SECs (Fig. 4C and Supplementary Fig. 4). Despite the increase of genetic alterations identified in the While 26.8% of the somatic mutations were common between the targeted validation of EEC7, the number of shared variants was five samples of SEC1, only 1% (54/6,700) of the genes affected by low in both the WES and the targeted sequencing (5.8% and SCNAs were shared across the six samples analyzed by aCGH. The 14.1%, respectively: Supplementary Fig. 3A). However, these three primary tumor regions harbored 87% (5856/6700) of the results provided evidence of the tumor evolution, since the SCNAs identified, yet only 2% (131/6700) were detected in the metastatic regions appeared to be developed from a common three metastatic regions, indicating higher levels of heterogeneity ancestral clone with the primary tumor, displaying private at the SCNA level in the primary tumor regions than in metastases metastatic mutations such as those in the RUNX1, ESR1 and JAK1 (Fig. 4C). It is worth mentioning the higher number of SCNAs genes (20% of the validated variants were exclusively identified in detected in SEC2 (11,838), being 39% of SCNAs identified in one of the metastatic regions, from M to M5). Thus, it can be the primary tumor regions analyzed (SEC2, T4). Increased ITGH hypothesized that metastatic lesions were dormant whereas the was observed at the SCNA level in SEC2, where only 5% of the primary tumor continued to evolve, or a possible clonal selection alterations identified were common between all the tumor after primary tumor treatment occurred, in accordance with the regions. Moreover, the metastasis and primary tumor tissue had phylogenetic tree (Fig. 3B bottom). In fact, 43.3% of the validated a high number of SCNAs, with 15.5% and 46.7% of the alterations variants were exclusively identified in the primary tumor regions observed, respectively (Fig. 4D). Similar results were observed for (T1, T2 and T3), suggesting that they were irrelevant for its SEC3, where 7.4% (364/4914) of the SCNAs detected were metastatic dissemination. Among the most remarkable variations common in all the regions analyzed and although the primary observed, the primary tumor regions carried mutations in ARID1A tumor regions had few SCNAs (2.8%), the metastatic regions and an indel in PIK3CA (p. E453del: Fig. 3B, Supplementary Table contained 15.5% of the SCNAs identified (Fig. 4E). S4). Thus, SCNAs were not uniformly distributed in the regions of The mutational profile of the 12 samples analyzed from the MSI the SEC tumors analyzed, suggesting the presence of ITGH in each case ECC6 was very consistent across. Common variants of PTEN, tumor. Mutations in TP53 tumor suppressor gene and/or ARID1A, KRAS and CTCF were identified in all regions analyzed, stabilization of the p53 protein are the most frequent alterations which shared 64.1% of the variants analyzed by targeted in SECs [31]. Although SEC3 did not harbor a TP53 mutation (Fig. sequencing (Supplementary Fig. 3A, Supplementary Table S4), 1C), we observed TP53 deletions in 4/5 SEC3 regions analyzed (Fig. although more heterogeneity was found among the primary 4F), consistent with the aberrant p53. Moreover, TP53 deletions tumor samples from different regions. This uniformity was were detected in some of the tumor regions of cases SEC1 and consistent with the low levels of ITGH observed among the tumor SEC2 analyzed. In addition to TP53 alterations, other focal regions analyzed (Fig. 3C). In this case, two main branches were deletions in FBXW7, FGFR1 and IGF1R genes were detected found in the phylogenetic tree, distinguishing between the [26, 32, 33]. Amplifications of the MYC, ERBB2, CCNE1 and PIK3CA primary tumor and metastatic regions in a monophyletic oncogenes, and of other genes like FGFR3 and SOX17 were also progression of the tumor (Fig. 3C bottom). observed in some EC samples (Fig. 4F), as previously reported in Low ITGH was observed for CN-high SECs (Fig. 4 and ECs [26]. Of note, the ERBB2 (HER2) amplification was validated in Supplementary Fig. 3B) and the CN-low ECs (Supplementary SEC1 at the protein level, in which HER2 membranous expression Fig. 3C, D). Most genetic variants in the SECs were shared by IHC (2+) was seen in the primary tumor region, being even consistently among the majority of the samples of a given case. more intense in both the metastatic sections analyzed (Fig. 4G), We noted, however, that the remaining mutations detected suggesting anti-HER2 therapy as a potential treatment option in were mainly identified in the primary tumor regions, suggesting this patient. Oncogene (2022) 41:1835 – 1850 A. Mota et al. 1841 Fig. 3 Targeted sequencing validation for MSI endometrial carcinomas. Validation of the pathogenic somatic mutations identified by targeted sequencing in primary tumor regions (T) and metastatic regions (M) of A EEC5, B EEC7 and C EEC6. The mutation subtypes are colored according to the legend and the origin of the tumor sections is indicated in the graphical representation: green line represents the peritoneum and orange line the diaphragm. Below each case the phylogenetic trees generated according to the presence or absence of the somatic mutations detected in the validation analysis are shown. MSI microsatellite instability, EEC endometrioid endometrial carcinoma. Oncogene (2022) 41:1835 – 1850 A. Mota et al. 1842 AEC characterization for personalized treatment differentiation (endometrioid, serous, clear cell), or features of WES revealed ATM mutations in all AEC tumor regions analyzed. undifferentiated carcinoma (Fig. 6A, Supplementary Fig. 1A). The Finally, we focused on the genomic profile of the AEC WES study of this tumor identified molecular features completely characterized by a solid arrangement and lack of any kind of different to the other EC samples. Despite the identification of 85 Oncogene (2022) 41:1835 – 1850 A. Mota et al. 1843 Fig. 4 Targeted validation and Copy Number Aberrations for serous endometrial carcinomas. Validation of pathogenic somatic mutations identified by targeted sequencing in primary tumor regions (T) and metastatic regions (M) of A SEC1 and B SEC2. The mutation subtypes are colored according to the legend and the origin of the tumor sections is indicated in the graphical representation. The phylogenetic tree generated according to the presence or absence of the somatic mutations detected in the validation analysis for each case is represented below. Dot graphs represent the proportion of copy number aberrations shared between primary tumor sections (T) and the metastatic regions (M) analyzed by aCGH of C SEC1, D SEC2 and E SEC3. F The status of the most representative genes with Copy Number Aberrations described in SECs are represented: red indicates amplification and green deletion. G Immunohistochemical staining for ERBB2 in the SEC1 primary tumor (T3) and metastatic (M2 and M4) sections. SEC serous endometrial carcinoma. Magnification ×20. Fig. 5 Analysis of the ambiguous endometrial carcinoma and personalized treatment. A Venn diagram representing the genetic variant distribution in two areas of the primary tumor (T1 and T2) and in the lymph node metastasis (M) of the ambiguous endometrial carcinoma (AEC) identified by whole-exome sequencing (WES) analysis. The numbers of genetic variants are shown with their percentage in brackets. B Phylogenetic tree based on somatic mutations depicting the evolution of the primary tumor areas and the metastatic regions, and a representation of the molecular signature obtained by the whole-exome sequencing (WES) (top) or targeted validation (bottom), color coded according to the legend. C Mutation Allele Frequency (MAF) heat-map based on the targeted sequencing validation of somatic mutations in multiple primary and metastatic samples from the ambiguous endometrial cancer and patient-derived xenograft (PDX) models. Tumor locations: T1–T5 different uterine locations of the primary tumor, M lymph node metastasis—detected and resected 7 years after primary tumor diagnosis, UA uterine aspirate obtained at time of the surgery. likely pathogenic somatic variants (Supplementary Table S1), none (Fig. 5B) similar to the majority of ECs was observed, although its of these involved genes were commonly altered in EC and SEC mutational profile was distinct from that of the remaining ECs subtypes. Instead, we identified a pathogenic mutation in all AEC analyzed. Despite the ambiguous features identified in the AEC case, tumor regions analyzed which affected a splice site of the Ataxia the patient showed reduced survival rates, similar to SEC (Supple- Telangiectasia Mutated gene (ATM, c.6096-1 G > A) and was mentary Fig. 1D). Interestingly, mutational signatures associated with associated with loss-of-heterozygosity (LOH). the AID/APOBEC family of cytidine deaminases (i.e., signatures 2 and The proportion of somatic mutations shared between the two 13 [34]: were detected in all AECs lesions analyzed. The APOBEC primary and the metastatic areas was high, with more than 63% of mutational signature has mainly been reported in bladder, breast, the genetic alterations being present in all the samples analyzed (Fig. cervical and head and neck cancers [28, 35–39], but also defects in 5A). The phylogenetic tree based on the somatic mutations had two DNA repair mechanisms including APOBEC have been described as a main branches, one corresponding to the primary tumor samples potential players in the progression from primary to metastasis in EC and the other to the metastatic sample. A monophyletic evolution [40, 41]. Oncogene (2022) 41:1835 – 1850 A. Mota et al. 1844 Targeted validation of the AEC, including uterine aspirates Olaparib as a potential treatment for the AEC patient. Finally, to and PDX identify new therapeutic approaches for the AEC patient, we used Like the other ECs, targeted sequencing was used to validate the molecular profile obtained for this patient to carry out an in the AEC genetic variants found in the WES study, validating 86 silico analysis in different drug databases. The results produced of 215 variants tested (40%) by Ampliseq sequencing (Supple- high scores for drugs classified as inhibitors of nucleic acid mentary Table S4 and Fig. 5B bottom). In addition to the synthesis, such as poly (ADP) ribose polymerase (PARP) inhibitors samples analyzed by WES, three FFPE areas of the primary [43] (Supplementary Table S6) and drugs like Bortezomib and tumor (T3, T4 and T5), as well as a uterine aspirate (UA) Paclitaxel. All samples analyzed by WES harbored pathogenic obtained at time of the surgery, were also analyzed (Supple- mutations in the ATM gene associated with LOH (Supplementary mentary Table S3). In the validation process, only 17% of the Table S1). In this regard, there have been reports of clinical variants were found to be common between all samples (T1-T5, benefits when the PARP inhibitor Olaparib has been used to treat M and UA) compared to 69% of the selected variants from WES cancer patients with somatic ATM mutations [44–46]. Thus, we (Supplementary Fig. S3A). Despite the differences observed assessed the benefits of treatment with Olaparib and Bortezomib among the tumor regions, metastasis, and UA, T1 and T2 shared relative to that of the standard EC treatment, carboplatin- the 96.4% of the mutations analyzed and T1-T3 and T2-T3 the paclitaxel [47, 48], in the F2 generation of the AEC_PDX1. A 69.4% for each case (Fig. 5C). However, this proportion was significant therapeutic effect was attributed to the Olaparib lower (<47%) between these three regions and T4 or T5, which treatment, which reduced tumor growth relative to the placebo could be related to the purity of primary T4 and T5 regions [42]. (p-value < 0.05: Fig. 6C). Together, these data support the This patient’s tumor had a polyphyletic evolution (Fig. 5B), feasibility of using genomic studies to identify new therapeutic whereby all the samples seemed to evolve from a single truncal opportunities, especially in cases of aggressive and/or ambiguous clone, with the primary T1 and T2 tumor regions, and the M tumors. lesion having a longer evolutionary process than T3, T4 and T5, as reflected by an accumulation of mutations restricted to these samples. These results were reflected in the large percentage DISCUSSION (77.9%) of mutations shared among T1, T2 and M. Furthermore, ITGH is a consequence of deficient DNA replication that has been 82% of the variants analyzed were detected in the UA sample, described since the early days of cancer research [49–51]. Genetic which also contained 22.8% of the somatic mutations detected differences not only exist between diverse cancers but also, within in the metastatic region and those were not found in T1, T2 and the same tumor [52]. The identification of cancer-driver mutations T3 (T4 and T5 were discarded due to low purity: Fig. 5C). and genomic alterations in tumors from different patients, and Although low levels of metastatic mutations were detected in within individual tumors, establishes the basis for precision the UA, this sample has the capacity to capture some of them in medicine, driving the treatment of the patient according to the a non-invasive way and recapitulate an important part of the genetic alterations present in their tumors. The development of ITGH in EC [13]. high-throughput techniques and specifically, massively parallel In parallel, we developed PDXs from the AEC as a preclinical sequencing, has made a real difference in our understanding model to gain insights into tumor evolution and the impact of of ITGH. therapies. A total of 5 different PDXs were set up from the AEC Numerous studies have shed light on the extent of tumor by implanting five different tumor samples into distinct mice, diversity in distinct solid tumor types, such as pancreatic [53, 54], samples from: the superficial region of the tumor (AEC_PDX1); lung [55, 56], breast [57–59], colorectal [60, 61] and prostate the deepest part of the tumor at the myometrial invasion front tumors [62, 63]. Similar analyses have been also carried out on (AEC_PDX2); the right lymphatic node metastasis (AEC_PDX3); gynecological cancers [64], mostly focusing on ovarian carcino- the left lymphatic node metastasis (AEC_PDX4); and the tumor mas [65, 66]. However, there are other tumor subtypes that have infiltrating the cervix (AEC_PDX5). After tumor growth, the PDXs been less well explored, as it is the case of EC. Here, at least two were resected and analyzed with a panel of biomarkers regions from primary EC tumors and one from a metastatic lesion previously used to differentiate between high-grade endome- were analyzed by WES, revealing that less than half of the trioid and serous carcinoma [24]. As previously observed in the mutations were shared between multiple samples from each EC AEC patient, the panel of biomarkers employed for the analysis patient, in line with results obtained for other tumor types of the PDXs (two PDXs showed as a representation in Fig. 6B [67, 68]. However, this differs from the homogeneity found in and Supplementary Fig. 5A) showed features from both pancreatic or lung carcinomas, where around 70% of the variants histological subtypes, with some biomarkers indicating a serous are ubiquitously detected [69–71]. For tumors that can be pattern while others an endometrioid one. The five PDX tumor completely resected, the sequencing of a single region is samples were analyzed by targeted sequencing, which generally adequate to select targeted therapies, however multi- detected 79 of the 86 variants analyzed for the AEC patient region sequencing is critical to evaluate the potential targeted- (81%: Fig. 5B bottom). In addition, a WES analysis was also therapies that could be effective when tumors cannot be performed on the PDXs and in normal mouse tissue as a control, completely removed and therefore, the majority of mutations identifying a total of 395 variants (Supplementary Table S5). are unlikely to be represented [52]. Interestingly, only 175 of the 395 (44%) variants had previously Here, different degrees of ITGH were found between the been identified in the patient’s samples, most of them were multiple EC molecular subgroups, the highest number of common between all the PDX samples (166/175, 95%). A total mutations identified for the MSI subgroup, coupled with the of 83 of the 258 variants (32%) previously detected in the highest level of ITGH. This ITGH was less pronounced for the CN- patient’s WES analysis were not identified in the PDXs high/SEC subgroup (30% common SNVs) followed by the AEC (Supplementary Fig. 5B). These findings suggested that PDXs patient (40% common SNV), being the lowest levels of ITGH found harbored most of the variants previously identified in the in the CN-low ECs, where more than 60% of the variants were respective human tumor, primarily those that were present in shared among all the regions. Furthermore, the majority of all the patient samples. Accordingly, the PDX models were variants detected were associated with a clonal status in each potentially derived from sub-clones of the tumor of origin, sample, suggesting that most of the cancer cells carried the which might underrepresent the ITGH found in the different majority of variants detected in this specific region when a single samples of the original tumor, and indeed, the PDXs may sample was considered. This could be misunderstood as the undergo additional evolution following implantation. absence of heterogeneity, however, clonal mutations in one Oncogene (2022) 41:1835 – 1850 A. Mota et al. 1845 Fig. 6 Histological and immunohistochemical profiles for the ambiguous endometrial carcinoma. Histological (HE) and immunophenotype of the ambiguous endometrial carcinoma based on a set of proteins described previously to help discriminate between SEC and differentiated EECs: MSH6, MLH1, MSH2, PMS2, PTEN, HMGA2, ER, ki67, IMP2, IMP3, CYC E1 and p53 [24] in AEC patient (A) and PDX derived from tumor 1 (B). Magnification ×20. C Growth of AEC-derived PDX tumors treated with the drugs indicated for 30 days. Tumor volumes were measured three times each week with a digital caliper and the volumes were calculated as: (length × width2)/2 = mm3; *p-value < 0.05, significantly different placebo vs. Carboplatin-Paclitaxel; +p-value < 0.05, significantly different placebo vs. Olaparib analyzed by one-way ANOVA. The data show the mean of 6–7 independent mice for each treatment. region of the tumor could be completely absent in another tumor The low sequencing depth of the WES and the low tumor region, as reported previously for renal [10], lung [70] or ovarian proportion sampled represent the most important limitations in [65] cancers. Regarding tumor progression, a high proportion of our study. WES studies provide the mutational landscape of the EC monophyletic evolution has been described in EC [27]. Indeed, we samples, the evolution of the regions analyzed and identify the observed monophyly in 7 of the 9 patients and polyphyletic molecular subtypes. Although our WES study provided evidence evolution in the remaining two cases (EEC5 and SEC2), although of the presence of ITGH in EC patients, there was more limited this earlier study did not examine multiple regions from the same detection of low-frequency mutations which often include private primary tumor [27]. mutations that represent the majority of the ITGH within a cancer Oncogene (2022) 41:1835 – 1850 A. Mota et al. 1846 [18]. To address this issue, pathogenic/recurrent variants identified chromothripsis or important genomic changes could participate in the WES were further analyzed in the validation study, which in metastasis, as observed in other cancers [82]. ERBB2 amplifica- included additional tumor samples. Interestingly, as well as the tion was detected in the CGH study for SEC1, SEC2 (both the common mutations found in each tumor, more specific variants primary tumor and metastasis) and SEC3 (only in metastasis), were mainly identified in the primary tumor regions in comparison while the IHC analysis of different tumor regions reflects the ITGH with the metastatic lesions. in the patient and could reveal new targets for available It is interesting to compare the metastatic timeline of EC to treatments. different organs based on the phylogenetic trees. Peritoneal and Regarding the mutational profile of AEC, we determined and lymph node metastasis are expected to occur late in the process confirmed the absence of the most frequent molecular alterations of tumor dissemination, as observed in the trees for EEC3, EEC4 in EC by two different approaches. AEC analysis revealed a and EEC5. However, the timeline for ovarian metastasis is probably mutational landscape not closely related to either high-grade ECs more variable, in particular for EEC. There is a subset of patients or SECs. AEC is an uncommon type of cancer, previously reported with EEC and ovarian metastasis that have a very good prognosis. in the literature [5–7] and difficult to classify at the morphological In fact, historically patients in this type of tumors were interpreted and molecular level. The diagnosis of AEC is hard to made in the as having two independent primary tumors due to their indolent absence of an intense molecular analysis, as was performed in our behavior [72–76]. However, mutation analysis suggested that both case. AEC cases are occasionally placed in the spectrum of high- tumors have a clonal origin and thus, ovarian lesions should be grade endometrioid carcinomas, although they are very aggres- considered as metastases [77–80]. Several hypotheses have been sive, and do not fit perfectly in any TCGA subgroup. For these proposed to explain such a favorable prognosis. The most feasible reasons, it is important to be aware of the existence of this explanation is that in this group of patients, ovarian metastases unusual type of tumor. Lower levels of ITGH were detected in AEC, occur through fallopian tube migration at early stages of tumor in which there was a stronger similarity between the two primary progression. By contrast, endometrioid carcinomas with ovarian tumors and the metastatic lesion. These findings suggest a metastasis and a poor prognosis would disseminate though monophyletic evolution of this tumor, whereby the primary and lymphovascular invasion, and this metastatic spread would occur metastatic lesions arose from an ancestral clone. Interestingly, late in the process of tumor progression. Nevertheless, in our signatures associated with the AID/APOBEC family were asso- series of cases, patients with ovarian metastasis had a worse ciated with the regions analyzed in the AEC, although these prognosis and died from their disease irrespective of their signatures are less frequent in EC [40, 41]. However, these molecular subgroup (EEC6, SEC1 and SEC3). The phylogenetic signatures have been correlated with immune cell populations trees of these tumors had a similar form, in which metastasis modulating the immunological response [83–85]. occurred early in tumor progression. These results could reflect Despite the decreased ITGH observed in our WES studies, this how the metastatic disease appears from an initial clone of the was stronger when additional tumor regions were studied by primary tumor with specific molecular features that then spreads targeted sequencing, highlighting the need for including different to the ovary, while the same tumor clone in the uterus would tumor sections to better characterize the tumor at the molecular continue acquiring mutations over time, modifying the initial level. Remarkably, our results confirm that UAs could recapitulate tumor profile. Nevertheless, the number of samples with this the ITGH in EC in a best way than traditional biopsies [13]. pattern analyzed in our study is limited and more samples should Furthermore, all the samples from the AEC patient analyzed by be included to extract definitive conclusions. A better prognosis WES had only one pathogenic variant that affected a splice and the same tumor evolution were observed for SEC2, which also acceptor site in the ATM gene, as validated by massive parallel developed an ovarian metastasis, and for the ECC7 with a targeted sequencing. This gene is involved in DNA repair and diaphragm metastasis, being both patients alive and disease-free. commonly found mutated in cancers associated with poor Recognition that ovarian metastasis occurs early in tumor outcomes and resistance to chemotherapy [86]. Consistent with dynamics in a subset of our patients supports the idea that the previous observations demonstrating that ATM and TP53 loss of timeline of metastasis may be related to the prognosis in these function mutations are mutually exclusive [87], this high-grade patients. However, additional tumors should be analyzed in order AEC lacked TP53 mutations, yet it harbored bi-allelic ATM to determine if this pattern might be related to tumor histology or inactivation. To further investigate the biology of this AEC, five to metastatic tropism. PDXs were set up from the superficial and deepest primary tumor Conversely, EEC3 and EEC4 patients had more variants in their areas, as well as from metastatic foci. Targeted massive parallel metastatic lesions, lymph nodes and omentum than in the sequencing revealed that most of the mutations previously primary tumor regions. In both cases, metastases were detected as detected in the patient (81%) were also found in all the PDXs a recurrent disease after 3 years of primary tumor treatment. analyzed, including that found in the ATM gene. This did not occur These results may indicate that recurrent metastatic samples with most of the variants that were heterogeneously identified in acquire additional mutations during or after tumor treatment, the patient, which were not observed in the PDX models. which favored their growth even after years of disease remission. Moreover, a personalized treatment for the AEC patient was In these cases, metastasis would slowly acquire a more complex explored, being PARP inhibitor considered a new potential genetic phenotype, which could lead to an aggressive behavior therapeutic approach in this specific tumor context. The PARP during their development. inhibitor Olaparib has a good response rate in patients with ATM Despite the low levels of ITGH observed in SECs, both in the mutations [44–46] and it significantly controlled tumor growth in WES and targeted sequencing studies, CGH studies highlighted the AEC PDX models. Overall, our findings suggest that Olaparib ITGH at the SCNA level among different tumor regions from the could be considered a potential treatment for this AEC patient. same patient for the most important genes affected by SCNA in EC Nevertheless, additional studies should evaluate if other EC [26]. In this molecular subgroup, amplifications and/or deletions patients with ATM mutations could benefit from therapy with were observed in the primary and metastatic regions, with neither Olaparib. significant differences among them nor clear evidence of Taken together, although our study presents some constraints metastasis-specific exome mutations, as reported previously for in the number of analyzed patients by subtype to extract outright EC and other cancers [27, 81]. Thus, the mutation pattern does not conclusions and more samples should be studied to resolve these fulfill an essential role in the development of metastasis, or there limitation, a mean of 9 sample regions per tumor have been are no common genetic mechanisms to generate them in ECs. It is analyzed which should be considered an important number to also important to note that other genomic alterations like infer the ITGH. Furthermore, this study emphasizes the Oncogene (2022) 41:1835 – 1850 A. Mota et al. 1847 Fig. 7 Summary of the procedure followed for samples included in the study. WES whole exome sequencing, P primary tumor, M metastasis, CNA copy number aberrations, PDX patient-derived xenografts. significance of analyzing the maximum number of samples to Biobank (record number B.0000745, ISCIII National Biobank Record), the better understand the tumor evolution, especially when the Xarxa Catalana de Bancs de Tumors and Plataforma de Biobancos ISCIII genetic heterogeneity found between multiple samples of the (PT13/0010/0014, B.000609: CEIC PR(AMI)364/2014). In addition, a uterine primary tumor is considered (Fig. 7). Nevertheless, it is not useful aspirate (UA), a minimally invasive sample of the endometrium collected for current management strategies, and it is fairly difficult to for diagnostic purposes, was obtained from the AEC patient and used in the validation study (Table 1). The patient samples were selected following calculate the number of tumor regions needed to ensure proper the study criteria, and the procedures established in current Spanish Law assessment of ITGH, not least as it is not possible to anticipate the 14/2007 on Biomedical Research and the 1716/2011 Royal Decree on degree of ITGH for each tumor. Another relevant issue is the Biobanks. The local ethical committee from each institution approved the sequencing methodology used in the ITGH study. Our study study and complete written informed consent was obtained from all the presents an important limitation of WES sequencing depth and in patients. The histological classification was performed by two pathologists the percentage of the cancer cell fraction (CCF), which could (ARS, XMG), determining the subtype according to the World Health overestimate the ITGH found. However, deeper validation Organization (WHO) criteria [90]. The International Federation of Gynecol- sequencing on a different platform and the use of additional ogy and Obstetrics (FIGO) guidelines [91] were used for the staging and grading of each tumor. samples can mitigate this problem and permit ITGH quantification. For each samples analysis DNA extraction, the Promise TCGA molecular Although several studies have suggested that ITGH could be an classification of EC patients, IHC analysis, Array Comparative Genome independent prognostic factor of disease progression and survival Hybridization (aCGH) methodology and TumorTracer Server have been in other types of tumors [88], the mechanisms underlying this detailed in Supplementary Material and Methods phenomenon remain unclear. The fact that some ECs show resistance to treatment has been related to the hypothetical ITGH Whole-exome sequencing of the primary tumor [89]. Indeed, reconstruction of the WES was performed in collaboration with Sistemas Genómicos, S.L. phylogenetic tree reflects how tumor evolution could be (Paterna, Valencia, Spain) and the MSKCC Integrated Genomics Operation correlated with tumor progression, prognosis, and the presence (NY, USA). Fresh-frozen regions were analyzed by WES using the of a recurrent disease (Fig. 7). Thus, our study takes into account SureSelectXT Human All Exon + UTR (71 Mb) v5 enrichment kit (Agilent, the relevance of studying different tumor regions by WES, for Santa Clara, CA) and paired-end sequencing was performed in an Illumina capturing the ITGH in EC patients with controversial clinicopatho- HiSeq 2500 sequencer (Illumina, San Diego, CA). A median depth of logical features, such as AEC, and could assist to spot the most coverage was achieved of 51x (range = 32–74×) for tumor samples and appropriate treatment set up. 52× (range = 45–60×) for normal samples. A total of 40 frozen tissue samples analyzed by WES were subjected to Ampliseq targeted re- sequencing to validate the variants selected. The Targeted resequencing analysis has been included in Supplementary Material and Methods. MATERIALS AND METHODS Human samples A series of 9 metastatic EC cases for which multiple fresh-frozen tumor Bioinformatics analysis zones were available were collected from the Arnau de Vilanova University The bioinformatics analyses were performed as described previously Hospital (Lleida), the MD Anderson Cancer Center (Madrid), the Vall [92, 93]. Briefly, reads from WES and targeted sequencing were aligned to d’Hebron University Hospital (Barcelona) and the Bellvitge University the human reference genome GRCh37 using Burrows-Wheeler Aligner [94]. Hospital (Barcelona), with the support of the MD Anderson Foundation Somatic single nucleotide variants (SNVs) were identified using MuTect Oncogene (2022) 41:1835 – 1850 A. Mota et al. 1848 [95], whereas small insertion and deletions (indels) were identified using 5. Soslow RA. Endometrial carcinomas with ambiguous features. Semin Diagn Strelka [96] and VarScan 2 [97], using paired tumor/normal approach to Pathol. 2010;27:261–73. exclude germline mutations. The potential functional effect of each non- 6. Espinosa I, D’Angelo E, Palacios J, Prat J. Mixed and ambiguous endometrial synonymous SNV was analyzed in silico as described previously [77], using carcinomas: a heterogenous group of tumors with different clinicopathologic a combination of multiple predictors. The CCF was calculated using and molecular genetic features. Am J Surg Pathol. 2016;40:972–81. ABSOLUTE and reviewed manually [26, 88, 98]. The mutational signature of 7. Garg K, Leitao MM, Wynveen CA, Sica GL, Shia J, Shi W, et al. p53 overexpression each EC studied was obtained by the analyzing the mutational context, as in morphologically ambiguous endometrial carcinomas correlates with adverse described previously [28, 77]. Briefly, SNVs were categorized into C > A, C > clinical outcomes. Mod Pathol. 2010;23:80–92. G, C > T, T > A, T > C or T > G and then sub-categorized according to the 8. Rabban JT, Gilks CB, Malpica A, Matias-Guiu X, Mittal K, Mutter GL, et al. Issues in nucleotides preceding (5') and succeeding (3') the mutated base. The the Differential Diagnosis of Uterine Low-grade Endometrioid Carcinoma, mutational patterns found in our samples were compared to the previously Including Mixed Endometrial Carcinomas: Recommendations from the Inter- published data [28], after normalization according to the frequency of national Society of Gynecological Pathologists. Int J Gynecol Pathol. 2019;38: nucleotide changes observed in the respective sequencing platforms. S25–S39. 9. Ding L, Ellis MJ, Li S, Larson DE, Chen K, Wallis JW, et al. Genome remodelling in a basal-like breast cancer metastasis and xenograft. Nature. 2010;464:999–1005. Phylogenetic tree generation and clonality study 10. Gerlinger M, Rowan AJ, Horswell S, Math M, Larkin J, Endesfelder D, et al. Maximum parsimony phylogenetic trees were generated using CCF values Intratumor heterogeneity and branched evolution revealed by multiregion from the WES analysis and the data on the presence/absence of somatic sequencing. N. Engl J Med. 2012;366:883–92. mutations in the validation of each tumor region, as described previously 11. Marusyk A, Almendro V, Polyak K. Intra-tumour heterogeneity: a looking glass [77]. Briefly, the R package Phangorn was used to apply the Neighbor- for cancer? Nat Rev Cancer. 2012;12:323–34. joining method and to obtain the Hamming distance, optimized using the 12. Yates LR, Gerstung M, Knappskog S, Desmedt C, Gundem G, Van Loo P, et al. parsimony Ratchet method [99]. To further determined the clonal Subclonal diversification of primary breast cancer revealed by multiregion composition of metastatic ECs in the WES study, a Bayesian clustering sequencing. Nat Med. 2015;21:751–9. model (PyClone [30]: was applied to variant allele fractions, incorporating 13. Mota A, Colas E, Garcia-Sanz P, Campoy I, Rojo-Sebastian A, Gatius S, et al. tumor cellularity, ploidy and local copy number obtained from ABSOLUTE Genetic analysis of uterine aspirates improves the diagnostic value and captures [98] and FACETS [100], as described previously [101]. the intra-tumor heterogeneity of endometrial cancers. Mod Pathol. 2017;30:134–45. In silico prediction of drug treatment 14. Casas-Arozamena C, Díaz E, Moiola CP, Alonso-Alconada L, Ferreirós A, Abalo A, Compound databases CTD (http://ctdbase.org/) and STITCH (http://stitch. et al. Genomic Profiling of Uterine Aspirates and cfDNA as an Integrative Liquid embl.de/: [102–104] were used to identify potentially effective drugs Biopsy Strategy in Endometrial Cancer. J Clin Med. 2020;9:585. according to the mutational status of genes identified in the WES carried 15. Ryu D, Joung JG, Kim NK, Kim KT, Park WY. Deciphering intratumor hetero- out on the AEC patient. The drugs identified were evaluated in a preclinical geneity using cancer genome analysis. Hum Genet. 2016;135:635–42. study using PDX models from the AEC patient (AEC_PDX). The significance 16. Gatius S, Cuevas D, Fernández C, Roman-Canal B, Adamoli V, Piulats JM, et al. was analyzed using the non-parametric Mann–Whitney test. Tumor heterogeneity in endometrial carcinoma: practical consequences. Pathobiology. 2018;85:35–40. 17. Jiang T, Shi W, Natowicz R, Ononye SN, Wali VB, Kluger Y, et al. Statistical Patient-derived xenografts: generation and treatment measures of transcriptional diversity capture genomic heterogeneity of cancer. PDXs were performed at the Biomedical Research Group in Gynaecology BMC Genomics. 2014;15:876. (Vall d’Hebron Institute of Research) and all the procedures involving 18. Shi W, Ng CKY, Lim RS, Jiang T, Kumar S, Li X, et al. Reliability of whole-exome animals were performed according to protocols approved by the Animal sequencing for assessing intratumor genetic heterogeneity. Cell Rep. Experimentation Ethics Committee at the Vall d’Hebron University Hospital 2018;25:1446–57. (CEEA 16/16). Details are specified in Supplementary Material and Methods. 19. Talhouk A, McConechy MK, Leung S, Yang W, Lum A, Senz J, et al. Confirmation of ProMisE: A simple, genomics-based clinical classifier for endometrial cancer. Statistical analysis Cancer. 2017;123:802–13. The data was compared between the TCGA dataset, and the patient series 20. Kommoss S, McConechy MK, Kommoss F, Leung S, Bunz A, Magrill J, et al. Final included in this project using the Mann–Whitney U test. Patient subgroups validation of the ProMisE molecular classifier for endometrial carcinoma in a were compared with an unpaired Student’s t-test (comparisons between large population-based case series. Ann Oncol. 2018;29:1180–8. two groups) or with one-way ANOVA and Tukey’s multiple comparison 21. Vermij L, Smit V, Nout R, Bosse T. Incorporation of molecular characteristics into tests (comparisons between more than two groups). Statistical analyses endometrial cancer management. Histopathology. 2020;76:52–63. were performed using the SPSS Statistics 17.0 software package (SPSS Inc., 22. Marquard AM, Birkbak NJ, Thomas CE, Favero F, Krzystanek M, Lefebvre C, et al. Chicago, IL). For SCNA comparison between WES and aCGH, “in house” TumorTracer: a method to identify the tissue of origin from the somatic python scripts were used. The concordance between both analyses was mutations of a tumor specimen. BMC Med Genomics. 2015;8:58. described with the kappa value using R [105]. OS was performed by 23. Rosa-Rosa JM, Leskelä S, Cristóbal-Lana E, Santón A, López-García M, Muñoz G, Kaplan–Meier statistics using follow-up data from EC patients to analyze et al. Molecular genetic heterogeneity in undifferentiated endometrial carci- the prognostic value of molecular/histological subgroups. OS studies were nomas. Mod Pathol. 2016;29:1594. performed using survival R package (https://CRAN.R-project.org/ 24. Santacana M, Maiques O, Valls J, Gatius S, Abó AI, López-García M, et al. A package=survival). 9-protein biomarker molecular signature for predicting histologic type in endometrial carcinoma by immunohistochemistry. Hum Pathol. 2014;45:2394–403. 25. Soslow RA, Tornos C, Park KJ, Malpica A, Matias-Guiu X, Oliva E, et al. Endo- DATA AVAILABILITY metrial Carcinoma Diagnosis: Use of FIGO Grading and Genomic Subcategories The datasets generated during the current study are available in the NCBI BioProject in Clinical Practice: Recommendations of the International Society of Gyneco- PRJNA378720 (ID: 378720) and with GEO accession number GSE101447. logical Pathologists. Int J Gynecol Pathol. 2019;38:S64–S74. 26. Cancer Genome Atlas Research N, Kandoth C, Schultz N, Cherniack AD, Akbani R, Liu Y, et al. Integrated genomic characterization of endometrial carcinoma. REFERENCES Nature. 2013;497:67–73. 1. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2020. CA Cancer J Clin. 27. Gibson WJ, Hoivik EA, Halle MK, Taylor-Weiner A, Cherniack AD, Berg A, et al. The 2020;70:7–30. genomic landscape and evolution of endometrial carcinoma progression and 2. Bokhman JV. Two pathogenetic types of endometrial carcinoma. Gynecol Oncol. abdominopelvic metastasis. Nat Genet. 2016;48:848–55. 1983;15:10–7. 28. Alexandrov LB, Nik-Zainal S, Wedge DC, Aparicio SA, Behjati S, Biankin AV, et al. 3. Lax SF, Kurman RJ. A dualistic model for endometrial carcinogenesis based on Signatures of mutational processes in human cancer. Nature. 2013;500:415–21. immunohistochemical and molecular genetic analyses. Verh Dtsch Ges Pathol. 29. Alexandrov LB, Kim J, Haradhvala NJ, Huang MN, Tian Ng AW, Wu Y, et al. The 1997;81:228–32. repertoire of mutational signatures in human cancer. Nature. 2020;578:94–101. 4. Kandoth C, Schultz N, Cherniack AD, Akbani R, Liu Y, Shen H, et al. Integrated 30. Roth A, Khattra J, Yap D, Wan A, Laks E, Biele J, et al. PyClone: statistical infer- genomic characterization of endometrial carcinoma. Nature 2013;497:67–73. ence of clonal population structure in cancer. Nat Methods. 2014;11:396–8. Oncogene (2022) 41:1835 – 1850 A. Mota et al. 1849 31. Bell DW, Ellenson LH. Molecular genetics of endometrial carcinoma. Annu Rev 57. Nik-Zainal S, Van Loo P, Wedge DC, Alexandrov LB, Greenman CD, Lau KW, et al. Pathol. 2019;14:339–67. The life history of 21 breast cancers. Cell. 2012;149:994–1007. 32. Kuhn E, Wu RC, Guan B, Wu G, Zhang J, Wang Y, et al. Identification of molecular 58. Demeulemeester J, Kumar P, Møller EK, Nord S, Wedge DC, Peterson A, et al. pathway aberrations in uterine serous carcinoma by genome-wide analyses. J Tracing the origin of disseminated tumor cells in breast cancer using single-cell Natl Cancer Inst. 2012;104:1503–13. sequencing. Genome Biol. 2016;17:250. 33. Le Gallo M, Bell DW. The emerging genomic landscape of endometrial cancer. 59. Rye IH, Trinh A, Saetersdal AB, Nebdal D, Lingjaerde OC, Almendro V, et al. Clin Chem. 2014;60:98–110. Intratumor heterogeneity defines treatment-resistant HER2+ breast tumors. Mol 34. Nik-Zainal S, Alexandrov LB, Wedge DC, Van Loo P, Greenman CD, Raine K, et al. Oncol. 2018;12:1838–55. Mutational processes molding the genomes of 21 breast cancers. Cell. 60. Wu H, Zhang XY, Hu Z, Hou Q, Zhang H, Li Y, et al. Evolution and heterogeneity 2012;149:979–93. of non-hereditary colorectal cancer revealed by single-cell exome sequencing. 35. Taylor BJ, Nik-Zainal S, Wu YL, Stebbings LA, Raine K, Campbell PJ, et al. DNA Oncogene. 2017;36:2857–67. deaminases induce break-associated mutation showers with implication of 61. Uchi R, Takahashi Y, Niida A, Shimamura T, Hirata H, Sugimachi K, et al. Cor- APOBEC3B and 3A in breast cancer kataegis. Elife. 2013;2:e00534. rection: integrated multiregional analysis proposing a new model of colorectal 36. Burns MB, Lackey L, Carpenter MA, Rathore A, Land AM, Leonard B, et al. cancer evolution. PLoS Genet. 2017;13:e1006798. APOBEC3B is an enzymatic source of mutation in breast cancer. Nature. 62. Gundem G, Van Loo P, Kremeyer B, Alexandrov LB, Tubio JMC, Papaemmanuil E, 2013;494:366–70. et al. The evolutionary history of lethal metastatic prostate cancer. Nature. 37. Robertson AG, Kim J, Al-Ahmadie H, Bellmunt J, Guo G, Cherniack AD, et al. 2015;520:353–7. Comprehensive molecular characterization of muscle-invasive bladder. Cancer 63. Kumar A, Coleman I, Morrissey C, Zhang X, True LD, Gulati R, et al. Substantial Cell. 2018;174:1033. interindividual and limited intraindividual genomic diversity among tumors 38. Cannataro VL, Gaffney SG, Sasaki T, Issaeva N, Grewal NKS, Grandis JR, et al. from men with metastatic prostate cancer. Nat Med. 2016;22:369–78. APOBEC-induced mutations and their cancer effect size in head and neck 64. Evans T, Matulonis U. Next-generation sequencing: role in gynecologic cancers. squamous cell carcinoma. Oncogene. 2019;38:3475–87. J Natl Compr Canc Netw. 2016;14:1165–73. 39. Zapatka M, Borozan I, Brewer DS, Iskar M, Grundhoff A, Alawi M, et al. The 65. Bashashati A, Ha G, Tone A, Ding J, Prentice LM, Roth A, et al. Distinct evolu- landscape of viral associations in human cancers. Nat Genet. 2020;52:320–30. tionary trajectories of primary high-grade serous ovarian cancers revealed 40. Ashley CW, Da Cruz Paula A, Kumar R, Mandelker D, Pei X, Riaz N, et al. Analysis through spatial mutational profiling. J Pathol. 2013;231:21–34. of mutational signatures in primary and metastatic endometrial cancer reveals 66. McPherson A, Roth A, Laks E, Masud T, Bashashati A, Zhang AW, et al. Divergent distinct patterns of DNA repair defects and shifts during tumor progression. modes of clonal spread and intraperitoneal mixing in high-grade serous ovarian Gynecol Oncol. 2019;152:11–9. cancer. Nat Genet. 2016;48:758–67. 41. Dessources K, Da Cruz Paula A, Pareja F, Stylianou A, Cybulska P, Farmanbar A, 67. Gerlinger M, Horswell S, Larkin J, Rowan AJ, Salm MP, Varela I, et al. Genomic et al. Acquisition of APOBEC Mutagenesis and Microsatellite Instability Sig- architecture and evolution of clear cell renal cell carcinomas defined by mul- natures in the Development of Brain Metastases in Low-Grade, Early-Stage tiregion sequencing. Nat Genet. 2014;46:225–33. Endometrioid Endometrial Carcinoma. JCO Precis Oncol. 2020;4:PO.20.00044. 68. McGranahan N, Swanton C. Clonal heterogeneity and tumor evolution: past, 42. Wilkerson MD, Cabanski CR, Sun W, Hoadley KA, Walter V, Mose LE, et al. present, and the future. Cell. 2017;168:613–28. Integrated RNA and DNA sequencing improves mutation detection in low purity 69. Yachida S, Jones S, Bozic I, Antal T, Leary R, Fu B, et al. Distant metastasis occurs tumors. Nucleic Acids Res. 2014;42:e107. late during the genetic evolution of pancreatic cancer. Nature. 2010;467:1114–7. 43. Peterson LE. Small molecule docking of DNA repair proteins associated with 70. de Bruin EC, McGranahan N, Mitter R, Salm M, Wedge DC, Yates L, et al. Spatial cancer survival following PCNA metagene adjustment: a potential novel class of and temporal diversity in genomic instability processes defines lung cancer repair inhibitors. Molecules. 2019;24:3. evolution. Science. 2014;346:251–6. 44. Zhang W, Shi J, Li R, Han Z, Li L, Li G. et al. Effectiveness of olaparib treatment in 71. Zhang J, Fujimoto J, Wedge DC, Song X, Seth S, Chow CW, et al. Intratumor a patient with gallbladder cancer with an ATM-inactivating mutation. Oncolo- heterogeneity in localized lung adenocarcinomas delineated by multiregion gist. 2020;25:375–9. sequencing. Science. 2014;346:256–9. 45. Schmitt A, Knittel G, Welcker D, Yang TP, George J, Nowak M, et al. Deficiency is 72. Connell PP, Rotmensch J, Waggoner S, Mundt AJ. The significance of adnexal associated with sensitivity to PARP1- and ATR inhibitors in lung adenocarci- involvement in endometrial carcinoma. Gynecol Oncol. 1999;74:74–9. noma. Cancer Res. 2017;77:3040–56. 73. Soliman PT, Slomovitz BM, Broaddus RR, Sun CC, Oh JC, Eifel PJ, et al. Syn- 46. Jette NR, Kumar M, Radhamani S, Arthur G, Goutam S, Yip S, et al. ATM- chronous primary cancers of the endometrium and ovary: a single institution deficient cancers provide new opportunities for precision oncology. Cancers. review of 84 cases. Gynecol Oncol. 2004;94:456–62. 2020;12:687. 74. Heitz F, Amant F, Fotopoulou C, Battista MJ, Wimberger P, Traut A, et al. Syn- 47. Kiess AP, Damast S, Makker V, Kollmeier MA, Gardner GJ, Aghajanian C, et al. chronous ovarian and endometrial cancer–an international multicenter case- Five-year outcomes of adjuvant carboplatin/paclitaxel chemotherapy and control study. Int J Gynecol Cancer. 2014;24:54–60. intravaginal radiation for stage I-II papillary serous endometrial cancer. Gynecol 75. Chiang YC, Chen CA, Huang CY, Hsieh CY, Cheng WF. Synchronous primary Oncol. 2012;127:321–5. cancers of the endometrium and ovary. Int J Gynecol Cancer. 2008;18:159–64. 48. Bestvina CM, Fleming GF. Chemotherapy for endometrial cancer in adjuvant 76. Irving JA, Catasús L, Gallardo A, Bussaglia E, Romero M, Matias-Guiu X, et al. and advanced disease settings. Oncologist. 2016;21:1250–9. Synchronous endometrioid carcinomas of the uterine corpus and ovary: 49. Heppner GH. Tumor heterogeneity. Cancer Res. 1984;44:2259–65. alterations in the beta-catenin (CTNNB1) pathway are associated with inde- 50. Gall JG. Human genome sequence. Science. 1986;233:1367–8. pendent primary tumors and favorable prognosis. Hum Pathol. 2005;36:605–19. 51. Hiley C, de Bruin EC, McGranahan N, Swanton C. Deciphering intratumor het- 77. Schultheis AM, Ng CK, De Filippo MR, Piscuoglio S, Macedo GS, Gatius S, et al. erogeneity and temporal acquisition of driver events to refine precision medi- Massively parallel sequencing-based clonality analysis of synchronous cine. Genome Biol. 2014;15:453. endometrioid endometrial and ovarian carcinomas. J Natl Cancer Inst. 52. Reiter JG, Baretti M, Gerold JM, Makohon-Moore AP, Daud A, Iacobuzio-Donahue 2016;108:djv427. CA, et al. An analysis of genetic heterogeneity in untreated cancers. Nat Rev 78. Anglesio MS, Wang YK, Maassen M, Horlings HM, Bashashati A, Senz J, et al. Cancer. 2019;19:639–50. Synchronous endometrial and ovarian carcinomas: evidence of clonality. J Natl 53. Campbell PJ, Yachida S, Mudie LJ, Stephens PJ, Pleasance ED, Stebbings LA, Cancer Inst. 2016;108:djv428. et al. The patterns and dynamics of genomic instability in metastatic pancreatic 79. Chao A, Wu RC, Jung SM, Lee YS, Chen SJ, Lu YL, et al. Implication of genomic cancer. Nature. 2010;467:1109–13. characterization in synchronous endometrial and ovarian cancers of endome- 54. Roberts NJ, Norris AL, Petersen GM, Bondy ML, Brand R, Gallinger S, et al. Whole trioid histology. Gynecol Oncol. 2016;143:60–7. genome sequencing defines the genetic heterogeneity of familial pancreatic 80. Reijnen C, Küsters-Vandevelde HVN, Ligtenberg MJL, Bulten J, Oosterwegel M, cancer. Cancer Discov. 2016;6:166–75. Snijders MPLM, et al. Molecular profiling identifies synchronous endometrial 55. Saber A, Hiltermann TJN, Kok K, Terpstra MM, de Lange K, Timens W, et al. and ovarian cancers as metastatic endometrial cancer with favorable clinical Mutation patterns in small cell and non-small cell lung cancer patients suggest a outcome. Int J Cancer. 2020;147:478–89. different level of heterogeneity between primary and metastatic tumors. Car- 81. Priestley P, Baber J, Lolkema MP, Steeghs N, de Bruijn E, Shale C, et al. Pan- cinogenesis. 2017;38:144–51. cancer whole-genome analyses of metastatic solid tumours. Nature. 56. Jamal-Hanjani M, Wilson GA, McGranahan N, Birkbak NJ, Watkins TBK, Veeriah S, 2019;575:210–6. et al. Tracking the evolution of non-small-cell lung cancer. N. Engl J Med. 82. Yi K, Ju YS. Patterns and mechanisms of structural variations in human cancer. 2017;376:2109–21. Exp Mol Med. 2018;50:98. Oncogene (2022) 41:1835 – 1850 A. Mota et al. 1850 83. Beck-Engeser GB, Winkelmann R, Wheeler ML, Shansab M, Yu P, Wünsche S, number B.0000745, ISCIII National Biobank Record), the “Xarxa Catalana de Bancs et al. APOBEC3 enzymes restrict marginal zone B cells. Eur J Immunol. de Tumors” and “Plataforma de Biobancos” ISCIII (PT13/0010/0014, B.000609). This 2015;45:695–704. study has been supported by the Spanish Ministry of Economy and Innovation 84. Norman JM, Mashiba M, McNamara LA, Onafuwa-Nuga A, Chiari-Fort E, Shen W, (PID2019-104644RB-I00 (GMB), the Instituto de Salud Carlos III (ISCIII, CIBERONC, et al. The antiviral factor APOBEC3G enhances the recognition of HIV-infected CB16/12/00295 - GMB-, CB16/12/00328 -EC, AGM- and CB16/12/00231 -XMG- [all primary T cells by natural killer cells. Nat Immunol. 2011;12:975–83. partly supported by FEDER funds]) and by the AECC Scientific Foundation 85. Budczies J, Seidel A, Christopoulos P, Endris V, Kloor M, Győrffy B, et al. Inte- (FC_AECC PROYE19036MOR -GMB- and Coordinated groups 2018 -XMG, AGM, grated analysis of the immunological and genetic status in and across cancer GMB-). SO is funded by an AECC-postdoctoral grant (2020). JSR-F and BW are types: impact of mutational signatures beyond tumor mutational burden. funded in part by the Breast Cancer Research Foundation and in part by the NIH/ Oncoimmunology. 2018;7:e1526613. NCI P50 CA247749 01 grant. Research reported in this publication was supported 86. Choi M, Kipps T, Kurzrock R. ATM mutations in cancer: therapeutic implications. in part by a Cancer Center Support Grant of the NIH/NCI (Grant No. P30CA008748; Mol Cancer Ther. 2016;15:1781–91. MSK). We thank the Eurofins Megalab laboratory for helping us to perform the 87. Weigelt B, Bi R, Kumar R, Blecua P, Mandelker DL, Geyer FC, et al. The landscape analysis of DNA HPV detection. of somatic genetic alterations in breast cancers from ATM germline mutation carriers. J Natl Cancer Inst. 2018;110:1030–4. 88. Landau DA, Carter SL, Stojanov P, McKenna A, Stevenson K, Lawrence MS, et al. AUTHOR CONTRIBUTIONS Evolution and impact of subclonal mutations in chronic lymphocytic leukemia. AM and SO performed the sequencing experiments, wrote the manuscript, and Cell. 2013;152:714–26. contributed to the analysis. BW, PS and JSR-F performed the bioinformatics and 89. Eritja N, Yeramian A, Chen BJ, Llobet-Navas D, Ortega E, Colas E, et al. Endo- statistical analysis. AM, ED and CC-A contributed to the sample collection and metrial carcinoma: specific targeted pathways. Adv Exp Med Biol. processing. AC was involved in the clinical follow-up of patients. MRM organized the 2017;943:149–207. samples and clinic pathological features. SG, AG-M, AR-S, AC and XM-G performed 90. Kurman R, Carcangiu M, Herrington C, Young R. WHO Classification of Tumours the pathological analysis of the samples. EC, CP-M, CL-G PH and RP contributed to the of Female Reproductive Organs. WHO Classification of Tumours. 4th Edition. pre-clinical models experiments. AM, SO, AG-M, JSR-F, XM-G, BW and GMB conceived Volume 6. World Health Organization, 2014. the study and participated in its design. AM, SO, XM-G, GMB, LM-R and MA helped to 91. Oncology FCoG. FIGO staging for carcinoma of the vulva, cervix, and corpus draft the manuscript. GMB discussed, directed the study and wrote the manuscript. uteri. Int J Gynaecol Obstet 2014;125:97–8. All the authors read and approved the final manuscript. 92. Weinreb I, Piscuoglio S, Martelotto LG, Waggott D, Ng CK, Perez-Ordonez B, et al. Hotspot activating PRKD1 somatic mutations in polymorphous low-grade ade- nocarcinomas of the salivary glands. Nat Genet. 2014;46:1166–9. 93. Guerini-Rocco E, Hodi Z, Piscuoglio S, Ng CK, Rakha EA, Schultheis AM, et al. The COMPETING INTERESTS repertoire of somatic genetic alterations of acinic cell carcinomas of the breast: JSR-F reports receiving personal/consultancy fees from Goldman Sachs, REPARE an exploratory, hypothesis-generating study. J Pathol. 2015;237:166–78. Therapeutics, Paige.AI and Eli Lilly, membership of the scientific advisory boards of 94. Li H, Durbin R. Fast and accurate short read alignment with Burrows-Wheeler VolitionRx, REPARE Therapeutics and Paige.AI, membership of the Board of Directors transform. Bioinformatics. 2009;25:1754–60. of Grupo Oncoclinicas, and ad hoc membership of the scientific advisory boards of 95. Cibulskis K, Lawrence MS, Carter SL, Sivachenko A, Jaffe D, Sougnez C, et al. Roche Tissue Diagnostics, Ventana Medical Systems, Novartis, Genentech and InVicro, Sensitive detection of somatic point mutations in impure and heterogeneous outside the scope of this study. BW reports ad hoc membership of the scientific cancer samples. Nat Biotechnol. 2013;31:213–9. advisory board of REPARE Therapeutics, outside the scope of this study. The 96. Saunders CT, Wong WS, Swamy S, Becq J, Murray LJ, Cheetham RK. Strelka: remaining authors have no conflict of interests to declare. accurate somatic small-variant calling from sequenced tumor-normal sample pairs. Bioinformatics. 2012;28:1811–7. 97. Koboldt DC, Zhang Q, Larson DE, Shen D, McLellan MD, Lin L, et al. VarScan 2: ADDITIONAL INFORMATION somatic mutation and copy number alteration discovery in cancer by exome Supplementary information The online version contains supplementary material sequencing. Genome Res. 2012;22:568–76. available at https://doi.org/10.1038/s41388-022-02221-0. 98. Carter SL, Cibulskis K, Helman E, McKenna A, Shen H, Zack T, et al. Absolute quantification of somatic DNA alterations in human cancer. Nat Biotechnol. Correspondence and requests for materials should be addressed to Gema Moreno- 2012;30:413–21. Bueno. 99. Schliep KP. phangorn: phylogenetic analysis in R. Bioinformatics. 2011;27:592–3. 100. Shen R, Seshan VE. FACETS: allele-specific copy number and clonal hetero- Reprints and permission information is available at http://www.nature.com/ geneity analysis tool for high-throughput DNA sequencing. Nucleic Acids Res. reprints 2016;44:e131. 101. Lee JY, Schizas M, Geyer FC, Selenica P, Piscuoglio S, Sakr RA, et al. Lobular Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims carcinomas in situ display intralesion genetic heterogeneity and clonal evolu- in published maps and institutional affiliations. tion in the progression to invasive lobular carcinoma. Clin Cancer Res. 2019;25:674–86. 102. Grondin CJ, Davis AP, Wiegers TC, King BL, Wiegers JA, Reif DM, et al. Advancing exposure science through chemical data curation and integration in the com- parative toxicogenomics database. Environ Health Perspect. 2016;124:1592–9. Open Access This article is licensed under a Creative Commons 103. Davis AP, Grondin CJ, Johnson RJ, Sciaky D, McMorran R, Wiegers J, et al. The Attribution 4.0 International License, which permits use, sharing, Comparative Toxicogenomics Database: update 2019. Nucleic Acids Res. adaptation, distribution and reproduction in any medium or format, as long as you give 2019;47:D948–D54. appropriate credit to the original author(s) and the source, provide a link to the Creative 104. Kuhn M, von Mering C, Campillos M, Jensen LJ, Bork P. STITCH: interaction Commons license, and indicate if changes were made. The images or other third party networks of chemicals and proteins. Nucleic Acids Res. 2008;36:D684–8. material in this article are included in the article’s Creative Commons license, unless 105. Tang W, Hu J, Zhang H, Wu P, He H. Kappa coefficient: a popular measure of indicated otherwise in a credit line to the material. If material is not included in the rater agreement. Shanghai Arch Psychiatry. 2015;27:62–7. article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons. ACKNOWLEDGEMENTS org/licenses/by/4.0/. We thank all those at the Translational Research Laboratory of the MD Anderson Cancer Center Madrid for their invaluable help with this study. Tissue samples © The Author(s) 2022 were obtained with the support of the MD Anderson Foundation Biobank (record Oncogene (2022) 41:1835 – 1850